Fructose-Bisphosphate Aldolase a Is a Key Regulator of Hypoxic Adaptation in Colorectal Cancer Cells and Involved in Treatment Resistance and Poor Prognosis

INTERNATIONAL JOURNAL OF ONCOLOGY 50: 525-534, 2017

Fructose-bisphosphate aldolase A is a key regulator of hypoxic adaptation in colorectal cancer cells and involved in treatment resistance and poor prognosis

Kenji Kawai1, Mamoru Uemura1,2, Koji Munakata1,3, Hidekazu Takahashi1, Naotsugu Haraguchi1, Junichi Nishimura1, Taishi Hata1, Chu Matsuda1, Masakazu Ikenaga4, Kohei Murata5, Tsunekazu Mizushima1,6, Hirofumi Yamamoto1,7, Yuichiro Doki1 and Masaki Mori1

1Department of Gastroenterological Surgery, Graduate School of Medicine, Osaka University, Suita, Osaka 565‑0871; 2Department of Surgery, National Hospital Organization Osaka National Hospital, Chuo-ku, Osaka, Osaka 540‑0006, Japan; 3Division of Gastroenterology, Department of Internal Medicine, University of Michigan, 3100A Taubman Center, Ann Arbor, MI 48109-5368, USA; 4Department of Surgery, Higashi-osaka City General Hospital, Higashiosaka, Osaka 578-8588; 5Department of Surgery, Suita Municipal Hospital, Suita, Osaka 564‑0082; Departments of 6Therapeutics for Inflammatory Bowel Diseases, and7 Molecular Pathology, Division of Health Sciences, Graduate School of Medicine, Osaka University, Osaka 565-0871, Japan

Received October 21, 2016; Accepted November 23, 2016

DOI: 10.3892/ijo.2016.3814

Abstract. Hypoxia is an essential feature of cancer malignancy, the expression of fructose-bisphosphate aldolase A (ALDOA) but there are no methods for the routine detection of hypoxia- and its relationship with cancer metabolism. We found that inducible prognostic factors and potential therapeutic targets. ALDOA was induced by hypoxia in CRC-derived cell lines, We reported previously that the hypoxic tumor cells of meta- and univariate and multivariate analyses of microarray data static liver tissue from patients with colorectal cancer (CRC) from the resected CRC samples of 222 patients revealed that could be used as an ‘in vivo’ hypoxia culture model. Several ALDOA was an independent prognostic factor for CRC. We potential hypoxia-inducible genes were identified using this also analyzed the malignant potential of ALDOA in vitro model. Among them, one glycolytic enzyme was of special using overexpression and knockdown assays. We found that interest. There is currently increasing attention on glycolytic ALDOA was negatively related to chemosensitivity and radio- enzymes as potential therapeutic targets due to their associa- sensitivity and positively associated with proliferation, sphere tion with cancer-specific metabolism. To better understand the formation and invasion in both normoxia and hypoxia. These molecular mechanisms of cancer malignancy, we investigated associations were due to the roles of ALDOA in regulating glycolysis, the epithelial-mesenchymal transition and the cell cycle. These findings demonstrate that ALDOA is a hypoxia- inducible prognostic factor that is closely related to CRC Correspondence to: Dr Mamoru Uemura, Department of Surgery, malignancy, and also provide new insights into the importance National Hospital Organization Osaka National Hospital, 2-1-14, of ALDOA and glycolysis in cancer and suggest new targets Houenzaka, Chuo-ku, Osaka 540-0006, Japan for anticancer therapies. E-mail: [email protected] Introduction Abbreviations: ADM, adrenomedullin; ALDOA, fructose-bis phosphate aldolase A; CA9, carbonic anhydrase IX; CRC, colorectal Colorectal cancer (CRC) is one of the most frequent cancers cancer; DFS, disease-free survival; EFNA1, ephrin-A1; EMT, in humans (1). Despite recent advances and new treatments, epithelial-mesenchymal transition; HIF1A, hypoxia-inducible factor 1-alpha; OS, overall survival; PCR, polymerase chain reaction; including combination therapy with anticancer and molecular PI, propidium iodide; PLOD2, procollagen-lysine, 2-oxoglutarate targeted drugs (2,3), CRC remains a major cause of cancer 5-dioxygenase 2; RIPA, radio immunoprecipitation assay; SCGB2A1, morbidity and mortality (4). Thus, there is still a need to iden- secretoglobin, family 2A, member 1; 5-FU, 5-fluorouracil tify new therapeutic targets in CRC. In the tumor microenvironment, cellular hyperplasia and Key words: fructose-bisphosphate aldolase A, chemoresistance, relative blood flow insufficiency often lead to tumor hypoxia. radioresistance, colorectal cancer, hypoxia To survive in this harsh environment, cancer cells develop adaptive responses to hypoxia that are mediated by the activation of hypoxia-inducible genes. In this context, hypoxia is a major feature of cancer, and in hypoxic conditions, cancer cells 526 KAWAI et al: ALDOA IS A KEY REGULATOR IN THE HYPOXIC ADAPTATION OF COLORECTAL CANCER can develop into malignant cells that survive in the low-oxygen MCO-5MUV; Panasonic, Osaka, Japan) using a continuous environment. Previous reports have suggested associations infusion of a gas mixture (95% N2, 5% CO2). between hypoxia and solid tumors, including brain tumors (5), breast (6), liver (7), lung 8), ovarian (9), prostate (10) and Ethics statement regarding the use of clinical samples. The CRC (11). use of clinical samples was approved by the Human Ethics Microarray techniques allow researchers to detect Review Committee of the Graduate School of Medicine, hypoxia-induced genes. Most hypoxia-related experiments Osaka University. Written informed consent was obtained are based on in vitro conditions. However, the problem with from all patients included in the study. hypoxic models is that there are apparent differences between the experimental hypoxic models and actual in vivo hypoxic Clinical tissue samples. For western blot analysis, 6 surgical environments. In experimental hypoxic conditions, there are CRC samples from patients with no preoperative chemo- many unpreventable biases, such as the degree of hypoxia, the therapy or irradiation were randomly selected from samples duration of the hypoxic conditions, pH, growth factors and collected at Osaka University Hospital in 2016. For immuno- nutrients. Moreover, it is almost impossible to replicate the histochemical staining, 14 surgical samples of colorectal liver complex cellular interactions that occur in vivo in an in vitro metastases from patients with no preoperative chemotherapy environment. or irradiation were collected from 2012 to 2015 at the Osaka We reported previously that liver metastasis of CRC is University Hospital. like ‘in vivo’ hypoxia culture box, and that it is useful for identifying hypoxia-inducible potential prognostic factors Clinical sample collection and microarray analysis. and therapeutic targets (12). We used this model to identify Microarray analysis was performed as previously described the following novel hypoxia-inducible genes: Jumonji domain (12,16,24). A total of 222 clinical samples were collected from containing 1A (JMJD1A) (12), adrenomedullin (ADM) (13), consecutive patients who underwent curative (R0) resection ephrin-A1 (EFNA1) (14), procollagen-lysine, 2-oxoglutarate for CRC from 2003 to 2006 at the Osaka University Hospital 5-dioxygenase 2 (PLOD2) (15) and secretoglobin, family 2A, and its nine associated hospitals. The mean follow-up time member 1 (SCGB2A1) (16). was 40.2±26.4 months for patients with disease-free survival In the present study, we investigated the involvement of (DFS) and 52.7±18.5 months for all surviving patients (overall fructose-bisphosphate aldolase A (ALDOA) in CRC. ALDOA survival, OS). Table I shows the clinicopathological features of is one of the glycolytic enzymes that catalyzes the reversible the patients, including gender, tumor location, depth of inva- conversion of fructose-1,6-bisphosphate to glyceraldehyde- sion, lymph node metastasis, histological grade, Dukes' stage 3-phosphate and dihydroxyacetone phosphate (17). Notably, and vessel invasion. None of the patients whose samples were previous studies have reported that ALDOA is a downstream analyzed in this study received preoperative chemotherapy or target of hypoxia-inducible factor 1-alpha (HIF1A) (18,19), and irradiation. After surgery, patients with Dukes' stage C tumors several studies have reported a relationship between ALDOA were generally treated with 5-fluorouracil (5-FU)-based and cancers such as osteosarcoma (20,21), lung squamous cell chemotherapy. carcinoma (22) and hepatocellular carcinoma (23). The aim of the present study was to investigate the prog- Quantitative PCR. mRNA expression was determined by real- nostic value and the biological significance of ALDOA in time polymerase chain reaction (PCR). Total RNA was isolated CRC. We determined the expression of ALDOA in clinical using an RNeasy Maxi kit (Qiagen, Valencia, CA, USA), and samples and performed functional analyses of ALDOA using reverse transcription was performed using the reverse tran- overexpression and knockdown assays. Furthermore, our data scription system (Promega, Madison, WI, USA) according to suggest possible mechanisms that underlie the correlation of the manufacturer's instructions. The real-time PCR analysis ALDOA with cancer malignancy. was performed using the LightCycler 2.0 instrument system (Roche Diagnostics, Mannheim, Germany), based on Materials and methods Thunderbird SYBR qPCR Mix (Toyobo Co., Ltd., Osaka, Japan). The β-actin housekeeping gene was used as an internal Cell lines and culture conditions. Six human CRC cell lines control. The following primers were used in the present study: were used in this study. HCT116 and LoVo cells were purchased ALDOA: forward 5'-CTGCCAGTATGTGACCGAGA-3' and from the Japanese Collection of Research Bioresources Cell reverse 5'-ACAGGAAGGTGATCCCAGTG-3'; HIF1A: Bank (JCRB; Tokyo, Japan), and HT29 and SW480 cells were forward 5'-GAAAGCGCAAGTCCTCAAAG-3' and reverse purchased from the American Type Culture Collection (ATCC; 5'-TGGGTAGGAGATGGAGATGC-3'; vimentin: forward Manassas, VA, USA). DLD-1 cells were purchased from the 5'-CCCTCACCTGTGAAGTGGAT-3' and reverse 5'-TCCA Cell Resource Center for the Biomedical Research Institute of GCAGCTTCCTGTAGGT-3'; and β-actin: forward 5'-GATG Development, Aging, and Cancer at Tohoku University. The AGATTGGCATGGCTTT-3' and reverse 5'-CACCTTCACC KM12SM cells were the kind gift of Professor T. Minamoto GTTCCAGTTT-3'. The analysis was performed using the (Kanazawa University, Ishikawa, Japan). Authenticity of all standard curve method. For data analysis, raw counts were cells was confirmed. Cells were cultured in Dulbecco's modi- normalized to the housekeeping gene average for the same fied Eagle's medium containing 10% fetal bovine serum (FBS), time-point and condition. 100 U/ml penicillin and 100 µg/ml streptomycin in a 37˚C incubator supplied with 5% CO2. Hypoxic culture conditions Western blotting. Proteins were extracted from the cells and (1% O 2) were achieved in a multi-gas incubator (Panasonic clinical samples in radio-immunoprecipitation assay (RIPA) INTERNATIONAL JOURNAL OF ONCOLOGY 50: 525-534, 2017 527

Table I. The relationship between ALDOA expression and clinicopathological factors in patients with colorectal cancer (n=222).

ALDOA expression ------Total (n=222) Low (n=111) High (n=111) P-value

Age (years); mean ± SD 66.2±10.0 65.3±9.6 67.1±10.2 0.24 Gender Male/female 133/89 65/46 68/43 0.78 Tumor location Colon/rectum 141/81 68/43 73/38 0.58 Depth of invasion ≤mp/≥ss 27/195 12/99 15/96 0.68 Lymph node metastasis Present/absent 109/113 45/66 64/47 0.016a Histological grade Well differentiated/other 53/169 27/84 26/85 1 Dukes' stage AB/CD 110/112 67/44 43/68 0.002a Vessel invasion Present/absent 184/38 89/22 95/16 0.37 mp, muscularis propria; ss, subserosa. ap<0.05.

buffer (Thermo Fisher Scientific, Waltham, MA, USA). MA, USA) and anti-CA9, 1:100 (#5649; Cell Signaling Clinical samples were homogenized in RIPA buffer using Technology). an ultrasonic disruptor (Ultrasonic Disruptor UD-201; Tomy Seiko, Co., Ltd., Tokyo, Japan). Proteins were separated Expression vector transfection. Cell lines were transfected by electrophoresis on SDS-PAGE Tris-HCl gels (Bio-Rad with an ALDOA expression vector (pCMV6-XL5 vector, Laboratories, Hercules, CA, USA). The separated proteins SC108418; Origene Technologies, Inc., Rockville, MD, were transferred onto polyvinylidene difluoride membranes USA) using the FuGENE®6 transfection reagent (Promega) and blocked with 5% skim milk in Tris-buffered saline with according to the manufacturer's recommended protocol. The Tween-20 (TBS-T) at room temperature for 1 h. The blots were same method was used to transfect cells with an empty vector incubated with primary antibodies overnight and then with as negative controls. HRP-linked anti-rabbit or anti-mouse IgG (GE Healthcare Biosciences, Piscataway, NJ, USA) at a dilution of 1:100,000 Small interfering RNA transfection. Cell lines were trans- for 1 h at room temperature. The antigen-antibody complex fected with 20 nM siRNA targeting ALDOA expression was detected with the ECL Prime Western blotting detection (Silencer Select siRNA, siRNA id s71, cat. no. 4390824; Life kit (GE Healthcare Biosciences). Technologies, Carlsbad, CA, USA: 5'-AGUCCCUCUUCGU CUCUAATT-3' and 5'-UUAGAGACGAAGAGGGACUAG-3') Antibodies for western blot analysis. Western blot analyses using Lipofectamine® RNAiMAX transfection reagent were carried out using the following antibodies: anti-ALDOA (Invitrogen, Carlsbad, CA, USA) according to the manufac- (#3188; Cell Signaling Technology, Danvers, MA, USA); turer's recommended protocol. The same method was used to anti-E-cadherin (sc-7870; Santa Cruz Biotechnology, Inc., transfect cells with negative control siRNA (Silencer Select Santa Cruz, CA, USA); anti-vimentin (#5741; Cell Signaling negative control #1 siRNA, cat. no. 4390843; Life Technologies). Technology) and anti-ACTB (A2066; Sigma-Aldrich, St. Louis, MO, USA). Proliferation assay. The proliferation assay was performed as previously described (16). ALDOA overexpression cells and Immunohistochemical staining. Paraffin-embedded tissue knockdown cells plus controls for each cell line were seeded sections (4-µm) were deparaffinized. After antigen retrieval and cultured in a 37˚C incubator supplied with 5% CO2. After treatment, immunohistochemical staining was performed 24, 48, 72 and 96 h of incubation, cell viability was measured using the Vectastain ABC Peroxidase kit (Vector Laboratories, using the Cell Counting kit-8 (CCK-8; Dojindo Laboratories, Burlingame, CA, USA). The tissue sections were then Kumamoto, Japan) according to the manufacturer's recom- incubated overnight at 4˚C with antibodies at the following mended protocol. CCK-8 solution was added to each well and dilutions: anti-ALDOA, 1:100 (ab169544; Abcam, Cambridge, incubated for 2 h. 528 KAWAI et al: ALDOA IS A KEY REGULATOR IN THE HYPOXIC ADAPTATION OF COLORECTAL CANCER

Chemosensitivity assay. The chemosensitivity assay was performed as previously described (16). Cell viability was measured after 48 h of incubation in drug-containing medium at IC50 concentrations.

Radiosensitivity assay. The radiosensitivity assay was performed as previously described (16). Cells were exposed to 8-Gy single-dose irradiation using a Gammacell 40 Exactor 137Cs radiation source (Nordion, Inc., Ottawa, ON, Canada) and cell viability was measured after 48 h of incubation in normal medium.

Sphere formation assay. The ability of cells to form spheres was evaluated in Dulbecco's modified Eagle's medium/ Nutrient Mixture F-12 (DMEM/F12) supplemented with 20 ng/ml epidermal growth factor (Invitrogen), 20 ng/ml human platelet growth factor (Sigma-Aldrich) and 1% antibi- otic-antimycotic solution (Invitrogen). Single cells were plated at a concentration of 100 cells/well in each well of a 96-well ultralow attachment plate (Corning Life Sciences, Acton, MA,

USA) and cultured in a 37˚C incubator supplied with 5% CO2. The number of spheres ≥100 µm in each well was counted to evaluate the sphere-forming ability.

Invasion assay. The invasion assay was performed using Corning® BioCoat™ Matrigel® Invasion Chambers (Corning Life Sciences). Briefly, 25,000 cells were seeded in triplicate onto the Matrigel-coated membrane. After 48 h, cells that had invaded the undersurface of the membrane were fixed and stained with Diff-Quik (Sysmex Internal Reagents, Co., Ltd., Kobe, Japan). Three microscopic fields were randomly selected for cell counting.

Lactate production assay. Lactate production was measured using the Lactate Colorimetric Assay kit II (BioVision, Inc., Milpitas, CA, USA). Cells were seeded in 12-well plates (Corning Life Sciences) at densities of 100,000 cells (DLD-1) and 150,000 cells (HT29 and KM12SM) per well. After 24 h, the cells were transfected with siRNA targeting ALDOA or with an ALDOA expression vector, and then 48 h after transfection, the subconfluent cells were washed three times with phosphate-buffered saline (PBS). Krebs-Ringer bicarbonate Figure 1. ALDOA expression is associated with cancer. (A) Western blot buffer containing 10 mM glucose was added, and the amount analysis of ALDOA protein expression in normal mucosa (N) and tumor of lactate was assessed 30 min later in aliquots of media from tissue (T) in 6 surgical samples. (B) Kaplan-Meier survival curves of the each well. The absorbance at 450 nm was then measured ALDOA high- and low-expression groups (log-rank test). (C) ALDOA mRNA expression in colorectal cancer (CRC) cell lines under hypoxia. using the Bio-Rad Model 680 XR microplate reader (Bio-Rad Bars represent the means ± SD of three independent experiments. *P<0.05, Laboratories). Student's t-test. (D) Representative images of immunohistochemical staining of colorectal cancer liver metastases. Scale bar, 100 µm. CA9, carbonic Apoptosis assay. Apoptosis was measured by flow cytometry anhydrase IX. using the Annexin V-FITC apoptosis kit (BioVision) according to the manufacturer's recommended protocol. The samples were analyzed using BD FACSAria IIu™ Fluorescence was measured using the BD FACSAria IIu instrument (BD Biosciences, San Jose, CA, USA). Both instrument (BD Biosciences). early apoptotic (Annexin V-FITC-positive, PI-negative) and necrotic/late apoptotic (Annexin V-FITC-positive, PI-positive) Cell cycle analysis. Cell cycle phase distribution was analyzed cells were quantified as apoptotic cells. by flow cytometry using PI staining. Briefly, cells were fixed in 70% ethanol on ice for 30 min. Fixed cells were incubated Analysis of reactive oxygen species (ROS). ROS production was in a solution containing 0.1 mg/ml RNase A at 37˚C for measured using the CellROX® Deep Red reagent (Invitrogen) 20 min and then in a solution containing 25 µg/ml PI on ice according to the manufacturer's recommended protocol. for 2 min. Fluorescence was measured immediately using INTERNATIONAL JOURNAL OF ONCOLOGY 50: 525-534, 2017 529

Table II. Univariate and multivariate analysis of clinicopathological factors associated with disease-free survival in patients with colorectal cancer (n=222).

Univariate analysis Multivariate analysis ------P-value Relative risk 95% CI P-value

Tumor location Colon/rectum 0.036a 0.592 0.288-0.908 0.0167a Depth of invasion ≤mp/≥ss 0.0058a 0.313 0.093-0.781 0.01a Lymph node metastasis Present/absent <0.0001a 1.324 0.852-2.100 0.2149 Histological grade Well differentiated/others 0.0193a 0.704 0.379-1.212 0.2147 Dukes' stage AB/CD <0.0001a Vessel invasion Present/absent 0.0002a 3.311 0.090-0.763 0.0088a ALDOA High/low 0.0008 2.01 1.310-3.127 0.0013a mp, muscularis propria; ss, subserosa; CI, confidence interval.a p<0.05.

the BD FACSAria IIu instrument. Flow cytometry data were low expression group than in the high expression group (DFS: analyzed using FlowJo software (Tree Star, Inc., Ashland, OR, P=0.0008, OS: P=0.0072; Fig. 1B). Univariate and multi- USA). variate analysis showed that ALDOA mRNA expression was a significant prognostic factor for DFS (P=0.0013; Table II) and Statistical analysis. Data management and statistical analysis for OS (P=0.0124; Table III). were performed using the JMP software (version 11.0; SAS Institute, Inc., Cary, NC USA). Tests for associations were ALDOA expression is upregulated under hypoxic conditions. performed using the Student's t-tests and chi-squared tests. In all of the CRC cell lines we examined, ALDOA mRNA The cumulative patient OS and DFS were compared using the expression was significantly increased after 48 to 72 h of culture Kaplan-Meier technique, and log-rank tests and Cox's model in hypoxic conditions compared with the levels in cells cultured were used to assess the risk ratio for multivariate analysis. A in normoxic conditions (Fig. 1C). Immunohistochemical P<0.05 was considered statistically significant. staining of CRC liver metastases showed that the regions stained by the anti-ALDOA antibody were almost the same as Results the regions stained by an antibody to carbonic anhydrase IX (CA9), which is an endogenous hypoxia marker (Fig. 1D). Relationship between ALDOA expression and clinicopathological features. Western blot analysis indicated that in 5 out of ALDOA expression correlates with proliferative potential 6 randomly selected CRC surgical samples, ALDOA expres- in CRC cells. PCR and western blot analysis demonstrated sion was higher in tumor tissue than in normal colon mucosa high transfer efficiency of the ALDOA expression vector (Fig. 1A). Then, the 222 CRC patients were divided into two and of the siRNA targeting ALDOA (data not shown). In groups according to the ALDOA mRNA expression level in the CRC-derived cell lines DLD-1 HT29 and KM12SM, the tumor samples. The relationships between ALDOA mRNA upregulation of ALDOA expression induced significant cell expression and clinicopathological features were analyzed proliferation compared to controls at 72 h (P<0.05; Fig. 2A), (Table I). The high ALDOA expression group included more and downregulation of ALDOA expression reduced cell patients with lymph node metastasis than the low expression proliferation compared to controls at 72 h (P<0.05; Fig. 2B). group (P=0.016), as for other clinicopathological factors, there were no significant differences. ALDOA confers chemoresistance and radioresistance. Upregulation of ALDOA significantly reduced chemosensi- ALDOA mRNA expression level is significantly associated tivity to 5-FU in DLD-1 and HT29 cells and to oxaliplatin with DFS and OS. DFS and OS were significantly longer in the in all three cell lines compared to controls (P<0.05; Fig. 2C). 530 KAWAI et al: ALDOA IS A KEY REGULATOR IN THE HYPOXIC ADAPTATION OF COLORECTAL CANCER

Figure 2. ALDOA is associated with the malignant potential of cancer. (A) The proliferation of cells transfected with the ALDOA expression vector. (B) The proliferation of cells transfected with siRNA targeting ALDOA expression. (C) The chemosensitivity and radiosensitivity of cells transfected with the ALDOA expression vector. (D) The chemosensitivity and radiosensitivity of cells transfected with siRNA targeting ALDOA expression. (E) Sphere formation by cells transfected with the ALDOA expression vector. Lower panel, representative images of a sphere; scale bar, 100 µm. (F) Sphere formation by cells transfected with siRNA targeting ALDOA expression. Lower panel, representative images of a sphere; scale bar, 100 µm. (G) Invasion by ALDOA overexpression cells (left) and ALDOA knockdown cells (right). Lower panel, representative images of the undersurface of the membrane; scale bar, 200 µm. The results represent the means ± SD of three independent experiments. 5-FU, 5-fluorouracil; KD, ALDOA knockdown; L-OHP, oxaliplatin; NC, negative control; OE, ALDOA overexpression. *P<0.05, Student's t-test.

In contrast, ALDOA downregulation significantly induced Correlation of ALDOA expression with sphere formation. chemosensitivity to 5-FU in DLD-1 and HT29 cells and DLD-1, HT29 and KM12SM cells that overexpressed ALDOA to oxaliplatin in DLD-1 and KM12SM cells compared to were more tumorigenic than controls (P<0.05; Fig. 2E). controls (P<0.05; Fig. 2D). Upregulation of ALDOA signifi- In DLD-1 and HT29, ALDOA knockdown cells were less cantly reduced radiosensitivity in DLD-1 and HT29 cells and tumorigenic than controls (P<0.05; Fig. 2F). that downregulation significantly induced radiosensitivity in DLD-1 and KM12SM cells compared to controls (P<0.05; ALDOA induces cell invasion. DLD-1 cells that overexpressed Fig. 2C and D). ALDOA invaded the Matrigel-coated membrane significantly INTERNATIONAL JOURNAL OF ONCOLOGY 50: 525-534, 2017 531

Table III. Univariate and multivariate analysis of clinicopathological factors of overall survival in patients with colorectal cancer (n=222).

Univariate analysis Multivariate analysis ------P-value Relative risk 95% CI P-value

Tumor location Colon/rectum 0.8715 Depth of invasion ≤mp/≥ss 0.241 Lymph node metastasis Present/absent <0.0001a 3.219 1.342-9.060 0.0075a Histological grade Well differentiated/others 0.0632 Dukes' stage AB/CD <0.0001a Vessel invasion Present/absent 0.0237a 1.635 0.444-10.579 0.498 ALDOA High/low 0.0072a 2.71 1.231-6.593 0.0124a mp, muscularis propria; ss, subserosa; CI, confidence interval.a p<0.05.

faster than control cells (P<0.05) and ALDOA knockdown chymal transition (EMT), we performed western blot analysis cells invaded significantly more slowly than control (P<0.05; to assess the expression of EMT markers in ALDOA knock- Fig. 2G). down DLD-1 and HT29 cells cultured under normoxic and hypoxic conditions. We found that E-cadherin expression was ALDOA expression is necessary for cancer cells to grow reduced by hypoxia and induced by knockdown of ALDOA in hypoxic conditions. To examine the role of ALDOA in both in normoxia and hypoxia. As for mesenchymal markers, hypoxia, the proliferation and sphere formation assay were vimentin expression level was too low to detect on western performed in hypoxic conditions. DLD-1 cells transfected with blots (Fig. 4A). PCR revealed that vimentin mRNA expression siRNA targeting ALDOA barely proliferated and formed very was reduced by ALDOA knockdown in both normoxia and few spheres compared to controls (P<0.05; Fig. 3A and B). hypoxia (P<0.05; Fig. 4B). Similarly, in the other cell lines, ALDOA knockdown cells formed hardly any spheres compared to controls (P<0.05; Knockdown of ALDOA stops cell cycle progression and Fig. 3B), collectively indicating that ALDOA is necessary enhances apoptosis due to rapid accumulation of ROS. The for proliferation and tumor formation in hypoxic conditions. rate of apoptosis was higher in hypoxic DLD-1 and HT29 cells In addition, PCR analysis showed that HIF1A expression is than in normoxic cells. It also showed that the rate of apoptosis upregulated by knockdown of ALDOA, probably because of a was higher in ALDOA knockdown cells than in controls in negative feedback loop (Fig. 3C). both hypoxia and normoxia (P<0.05; Fig. 5A). In addition, the ROS-positive cell rate of ALDOA knockdown cells was ALDOA expression is proportional to the activity of glycolytic higher than controls both in hypoxia and in normoxia (P<0.05; pathway. In DLD-1, HT29 and KM12SM cells, the knock- Fig. 5B). Cell cycle analysis showed that the G2/G1 ratio was down of ALDOA resulted in a decrease in lactate production higher in ALDOA knockdown cells than in controls in both compared to controls, and, in contrast, its overexpression hypoxia and normoxia (P<0.05; Fig. 5C). resulted in an increase in lactate production (P<0.05; Fig. 3D). This means that the expression of ALDOA and the activity of Discussion the glycolytic pathway were positively correlated. There has been increasing interest in glycolytic enzymes ALDOA expression is associated with EMT progression. as potential therapeutic targets because the cancer-specific ALDOA knockdown assay was performed under hypoxia to metabolism depends more on the glycolytic pathway than assess the function of ALDOA in hypoxic conditions. To focus on aerobic respiration, a phenomenon called the Warburg on the relationship between ALDOA and epithelial-mesen- effect (25). The present report is the first to characterize the 532 KAWAI et al: ALDOA IS A KEY REGULATOR IN THE HYPOXIC ADAPTATION OF COLORECTAL CANCER

Figure 4. ALDOA expression is associated with the epithelial-mesenchymal transition (EMT). (A) Western blot analysis of the EMT markers E-cadherin and vimentin in ALDOA knockdown cells in normoxic and hypoxic conditions. SW480 was used as a positive control for vimentin. (B) Vimentin mRNA expression in ALDOA knockdown cells in normoxic and hypoxic conditions. The results represent the means ± SD of three independent experiments. KD, ALDOA knockdown; NC, negative control. *P<0.05, Student's t-test.

of CRC, including cancer cell proliferation, sphere formation, chemoresistance, radioresistance and invasion. It seems likely that the relevance of ALDOA to cancer proliferation, sphere formation, and radioresistance is based on the function of ALDOA in glycolysis and cell cycle regulation. Other studies have reported that the glycolytic pathway and the expression of glycolytic enzymes are associated with cancer radiosensitivity (26,27). We showed that ALDOA expression is positively correlated with the activity of glycolysis, with cell cycle progression and with ROS production. ALDOA is an isozyme of the aldolase family, and a previous study showed that aldolase is involved in the cell cycle. Its involvement in Figure 3. ALDOA is required for cancer cells to grow in hypoxic condi- cell cycle progression is mediated by the binding of aldolase to tions, and the expression is proportional to the activity of glycolytic F-actin, with the cell cycle defects observed at M phase (28). pathway. (A) Proliferation of DLD-1 cells transfected with siRNA targeting Our flow cytometry data were consistent with these data. As ALDOA expression performed under hypoxia. (B) Sphere formation by cells transfected with siRNA targeting ALDOA expression cultured under a result of the cell cycle defect at M phase, the G2/G1 ratio hypoxia. (C) HIF1A mRNA expression by cells transfected with siRNA increased when ALDOA was knocked down. Both inhibition targeting ALDOA expression. (D) Aliquots of the medium from cultures of the glycolytic pathway and the arrest of cell division caused of ALDOA knockdown cells (left) and ALDOA overexpression cells (right) by ALDOA knockdown induced apoptosis, probably because were removed at the indicated time points and tested for lactate. The results represent the means ± SD of three independent experiments. HIF1A, of the increase in oxidative stress and the inhibition of tumor hypoxia-inducible factor 1-alpha; KD, ALDOA knockdown; LDHA, lac- progression. tate dehydrogenase A; NC, negative control; OE, ALDOA overexpression. As for the relevance of ALDOA in chemoresistance and *P<0.05, Student's t-test. invasion, we suggested an association of ALDOA expression with EMT progression. Other studies have previously concluded that the EMT correlates with chemoresistance importance of ALDOA in CRC and to investigate in detail the (29,30), thus, it appeared that the association of ALDOA possible mechanisms underlying that relationship. expression with the EMT led to chemoresistance and cancer The most meaningful finding in the present study was that invasion in CRC cell lines. This seems in line with the there is a close association between ALDOA expression and observed relationship between ALDOA expression and the the clinicopathological features and the malignant potential clinicopathological features of CRC i.e. the ALDOA high INTERNATIONAL JOURNAL OF ONCOLOGY 50: 525-534, 2017 533

Figure 5. ALDOA expression is associated with cell cycle and apoptosis. (A) Apoptosis in ALDOA knockdown cells in normoxic and hypoxic conditions. (B) Analysis of reactive oxygen species (ROS) activity in ALDOA knockdown cells in normoxic and hypoxic conditions. (C) Cell cycle analysis of ALDOA knockdown cells in normoxic and hypoxic conditions. The results represent the means ± SD of three independent experiments. KD, ALDOA knockdown; NC, negative control. *P<0.05, Student's t-test.

expression group included more patients with lymph node tion and sphere formation, but the gap with the control was metastasis than the low expression group (Table I). not so clear as in hypoxia. It is speculated that in hypoxic ALDOA is a hypoxia-inducible gene and a downstream conditions, cancer cells may try to adapt to hypoxia by target of HIF1A. Hypoxia is an important feature in cancer upregulating hypoxia-inducible genes; however, when malignancy, and under hypoxic conditions the glycolytic ALDOA is knocked down, they are unable to adapt, and the pathway is upregulated by activating HIF1A. Notably, the growth of cancer cells stops. This implies that particularly in EMT is upregulated in hypoxic environments (31), while cell hypoxic conditions, cancer cells could not grow well without cycle progression is inhibited (7); accordingly, the present ALDOA expression. study verified the effect of ALDOA on cancer under hypoxic In summary, we found that ALDOA positively regulated conditions. Our data revealed that especially under hypoxic the activity of the glycolytic pathway and the EMT and was conditions, ALDOA is essential for CRC cell proliferation necessary for cell cycle progression. Thus, ALDOA is impli- and sphere formation. On the other hand, knockdown of cated in CRC proliferation, sphere formation, therapeutic ALDOA in normoxic conditions resulted in less prolifera- resistance and invasion. 534 KAWAI et al: ALDOA IS A KEY REGULATOR IN THE HYPOXIC ADAPTATION OF COLORECTAL CANCER

Acknowledgements 15. Noda T, Yamamoto H, Takemasa I, Yamada D, Uemura M, Wada H, Kobayashi S, Marubashi S, Eguchi H, Tanemura M, et al: PLOD2 induced under hypoxia is a novel prognostic factor The present study was supported by the JSPS KAKENHI for hepatocellular carcinoma after curative resection. Liver Int Grand Number 15K10140 and the Takeda Science Foundation. 32: 110-118, 2012. 16. Munakata K, Uemura M, Takemasa I, Ozaki M, Konno M, Nishimura J, Hata T, Mizushima T, Haraguchi N, Noura S, et al: References SCGB2A1 is a novel prognostic marker for colorectal cancer associated with chemoresistance and radioresistance. Int J Oncol 1. Weitz J, Koch M, Debus J, Höhler T, Galle PR and Büchler MW: 44: 1521-1528, 2014. Colorectal cancer. Lancet 365: 153-165, 2005. 17. Tochio T, Tanaka H, Nakata S and Hosoya H: Fructose-1,6- 2. Beppu N, Yoshie H, Kimura F, Aihara T, Doi H, Kamikonya N, bisphosphate aldolase A is involved in HaCaT cell migration by inducing lamellipodia formation. J Dermatol Sci 58: 123-129, Matsubara N, Tomita N, Yanagi H and Yamanaka N: 2010. Clinicopathological outcomes of preoperative chemoradio- 18. Semenza GL, Roth PH, Fang HM and Wang GL: Transcriptional therapy using S-1 plus Irinotecan for T4 lower rectal cancer. Surg regulation of genes encoding glycolytic enzymes by hypoxia- Today 46: 852-859, 2016. inducible factor 1. J Biol Chem 269: 23757-23763, 1994. 3. Saeki H, Emi Y, Kumashiro R, Otsu H, Kawano H, Ando K, 19. Sharp FR and Bernaudin M: HIF1 and oxygen sensing in the Ida S, Kimura Y, Tokunaga E, Oki E, et al: Impact of second-line brain. Nat Rev Neurosci 5: 437-448, 2004. and later cetuximab-containing therapy and KRAS genotypes in 20. Chen X, Yang TT, Zhou Y, Wang W, Qiu XC, Gao J, Li CX, patients with metastatic colorectal cancer: A multicenter study in Long H, Ma BA, Ma Q, et al: Proteomic profiling of osteosar- Japan. Surg Today 44: 1457-1464, 2014. coma cells identifies ALDOA and SULT1A3 as negative survival 4. Jemal A, Siegel R, Xu J and Ward E: Cancer statistics, 2010. CA markers of human osteosarcoma. Mol Carcinog 53: 138-144, Cancer J Clin 60: 277-300, 2010. 2014. 5. Jensen RL: Brain tumor hypoxia: Tumorigenesis, angiogenesis, 21. Long F, Cai X, Luo W, Chen L and Li K: Role of aldolase A imaging, pseudoprogression, and as a therapeutic target. J in osteosarcoma progression and metastasis: In vitro and in vivo Neurooncol 92: 317-335, 2009. evidence. Oncol Rep 32: 2031-2037, 2014. 6. Vaupel P, Briest S and Höckel M: Hypoxia in breast cancer: 22. Du S, Guan Z, Hao L, Song Y, Wang L, Gong L, Liu L, Qi Pathogenesis, characterization and biological/therapeutic impli- X, Hou Z and Shao S: Fructose-bisphosphate aldolase A is a cations. Wien Med Wochenschr 152: 334-342, 2002. potential metastasis-associated marker of lung squamous cell 7. Zhang L, Huang G, Li X, Zhang Y, Jiang Y, Shen J, Liu J, carcinoma and promotes lung cell tumorigenesis and migration. Wang Q, Zhu J, Feng X, et al: Hypoxia induces epithelial-mesen- PLoS One 9: e85804, 2014. chymal transition via activation of SNAI1 by hypoxia-inducible 23. Shimizu T, Inoue K, Hachiya H, Shibuya N, Shimoda M and factor -1α in hepatocellular carcinoma. BMC Cancer 13: 1-9, Kubota K: Frequent alteration of the protein synthesis of 2013. enzymes for glucose metabolism in hepatocellular carcinomas. 8. Lee GW, Go SI, Cho YJ, Jeong YY, Kim HC, Duk Lee J, J Gastroenterol 49: 1324-1332, 2014. Hwang YS, Ko GH, Lee JH, Kim DC, et al: Hypoxia-inducible 24. Munakata K, Uemura M, Tanaka S, Kawai K, Kitahara T, factor-1α and excision repair cross-complementing 1 in patients Miyo M, Kano Y, Nishikawa S, Fukusumi T, Takahashi Y, et al: with small cell lung cancer who received front-line platinum- Cancer stem-like properties in colorectal cancer cells with low based chemotherapy: A retrospective study. J Thorac Oncol 7: proteasome activity. Clin Cancer Res 22: 5277-5286, 2016. 528-534, 2012. 25. Warburg O: On the origin of cancer cells. Science 123: 309-314, 9. Liang D, Ma Y, Liu J, Trope CG, Holm R, Nesland JM and Suo Z: 1956. The hypoxic microenvironment upgrades stem-like properties of 26. Meng MB, Wang HH, Guo WH, Wu ZQ, Zeng XL, Zaorsky NG, ovarian cancer cells. BMC Cancer 12: 201, 2012. Shi HS, Qian D, Niu ZM, Jiang B, et al: Targeting pyruvate 10. Higgins LH, Withers HG, Garbens A, Love HD, Magnoni L, kinase M2 contributes to radiosensitivity of non-small cell lung Hayward SW and Moyes CD: Hypoxia and the metabolic cancer cells in vitro and in vivo. Cancer Lett 356: 985-993, 2015. phenotype of prostate cancer cells. Biochim Biophys Acta 1787: 27. Pitroda SP, Wakim BT, Sood RF, Beveridge MG, Beckett MA, 1433-1443, 2009. MacDermed DM, Weichselbaum RR and Khodarev NN: STAT1- 11. Chang LH, Chen CH, Huang DY, Pai HC, Pan SL and Teng CM: dependent expression of energy metabolic pathways links tumour Thrombin induces expression of twist and cell motility via the growth and radioresistance to the Warburg effect. BMC Med 7: hypoxia-inducible factor-1α translational pathway in colorectal 68, 2009. cancer cells. J Cell Physiol 226: 1060-1068, 2011. 28. Ritterson Lew C and Tolan DR: Targeting of several glycolytic 12. Uemura M, Yamamoto H, Takemasa I, Mimori K, Hemmi H, enzymes using RNA interference reveals aldolase affects cancer Mizushima T, Ikeda M, Sekimoto M, Matsuura N, Doki Y, et al: cell proliferation through a non-glycolytic mechanism. J Biol Jumonji domain containing 1A is a novel prognostic marker for Chem 287: 42554-42563, 2012. colorectal cancer: In vivo identification from hypoxic tumor 29. Fischer KR, Durrans A, Lee S, Sheng J, Li F, Wong ST, Choi H, cells. Clin Cancer Res 16: 4636-4646, 2010. El Rayes T, Ryu S, Troeger J, et al: Epithelial-to-mesenchymal 13. Uemura M, Yamamoto H, Takemasa I, Mimori K, Mizushima T, transition is not required for lung metastasis but contributes to chemoresistance. Nature 527: 472-476, 2015. Ikeda M, Sekimoto M, Doki Y and Mori M: Hypoxia-inducible 30. Singh A and Settleman J: EMT, cancer stem cells and drug resis- adrenomedullin in colorectal cancer. Anticancer Res 31: 507-514, tance: An emerging axis of evil in the war on cancer. Oncogene 2011. 29: 4741-4751, 2010. 14. Yamamoto H, Tei M, Uemura M, Takemasa I, Uemura Y, 31. Majmundar AJ, Wong WJ and Simon MC: Hypoxia-inducible Murata K, Fukunaga M, Ohue M, Ohnishi T, Ikeda K, et al: factors and the response to hypoxic stress. Mol Cell 40: 294-309, Ephrin-A1 mRNA is associated with poor prognosis of colorectal 2010. cancer. Int J Oncol 42: 549-555, 2013.